I heard a great example of global misunderstanding yesterday in a discussion about testing at scale for “big data” type systems. It was claimed, and I accept for the purpose of this post, that you couldn’t test an application that took 5-hours on 20,000 servers, maybe so. However, if the data was considered personal information under most western economies privacy laws, that meant you couldn’t test against production data anyway. So you go live without testing or have to have a process to anonymise the data before you test on it. Then there’s the 20,000 server problem… So in the … Continue reading Test is production